Comparative study between neuromorphic implementations using Spiking Neural Network connected by Network-on-Chip

Authors

  • Victor Benvenutti Cavalheiro Universidade Federal de Santa Catarina
  • Beatriz Oliveira Câmara
  • Janaina Gonçalves Guimarães

DOI:

https://doi.org/10.5335/rbca.v17i1.16019

Keywords:

Hardware, Network-on-Chip, Neuromorphic Chips, PRISMA, Spiking Neural Networks

Abstract

The efficient implementation of neural networks in hardware, especially using networks-on-chip for the interconnection of spiking neural networks in neuromorphic chips, is a significant advancement in computational systems that mimic the human brain. This study conducts a systematic review on the integration of these technologies in neuromorphic chips, using the PRISMA methodology to analyze various studies and compare different hardware approaches. The results highlight the importance of energy-efficient and high-performance solutions for artificial intelligence. Despite technical challenges, neuromorphic computing is rapidly evolving, with the potential to advance several emerging technologies. The challenges include exploring neuron models, learning techniques, and interconnection and routing strategies to improve efficiency and performance.

Downloads

Download data is not yet available.

Published

2025-05-23

Issue

Section

Original Paper

How to Cite

[1]
2025. Comparative study between neuromorphic implementations using Spiking Neural Network connected by Network-on-Chip. Brazilian Journal of Applied Computing. 17, 1 (May 2025), 1–11. DOI:https://doi.org/10.5335/rbca.v17i1.16019.